15  Next Steps

What should you do next? You’ve learned some statistics, specifically

• numerical description
• visual description
• linear regression
• linear regression diagnostics

You’ve also been introduced to hypothesis testing, confidence intervals, and classification (through logistic regression).

Here are my ideas about topics you should learn next.

• Basic calculus through partial derivatives (the Calculus Tutoring Book is a good start)
• Probability (the Probability Tutoring Book is a good start)
• Linear algebra (there are dozens of free linear algebra books online) (although I’m partial to the nonfree linear algebra introduction in the Goodfellow Deep Learning book and the nonfree Howard Anton linear algebra textbook)
• Statistical learning (read the free ISLR2 book!)
• Bias-Variance Tradeoff (covered in the ISLR2 book)
• Transformer neural networks (for an intro on neural networks in general, see https://youtu.be/aircAruvnKk?si=zjHr3lJJ4vY91BaJ and the following videos) (covering MLPs multi layer perceptrons, backpropagation, stochastic gradient descent, RELU, and more)
• Read the paper “Attention is all you need”
• Read the paper “Textbooks are all you need”
• Read about retrieval augmented generation (RAG)
• Read about generative adversarial networks (GAN)

Then, after you’ve learned at least some of the above topics, fine-tune a small generative AI model using one of the pre-trained models you can find in the Hugging Face Hub repository. This will teach you much more about generative AI than you will from listening to others talk about it. But still, you won’t understand what you are doing unless you get some grounding in probability and linear algebra.

I realize that the above is a long list and that you probably won’t do everything on it! I’ve tried to put the most important things at the top in the sense that you won’t understand the items lower down on the list unless you understand probability and linear algebra, for instance. Those are foundational. The rest is an adventure!